MongoDB Data Model Optimization - performance

Question

About
I have raw data which is processed by aggregation framework and later results saved in another collection. Let's assume aggregation results in something like this:
cursor = {
"result" : [
{
"_id" : {
"x" : 1,
"version" : [
"2_0"
],
"date" : {
"year" : 2015,
"month" : 3,
"day" : 26
}
},
"x" : 1,
"count" : 2
}
],
"ok" : 1
};
Note that in most cases cursor length are more than about 2k elements.
So, now i'm looping thought cursor ( cursor.forEach ) and performing following steps:
// Try to increment values:
var inc = db.runCommand({
findAndModify: my_coll,
query : {
"_id.x" : "1",
"value.2_0" : {
"$elemMatch" : {
"date" : ISODate("2015-12-18T00:00:00Z")
}
}
},
update : { $inc: {
"value.2_0.$.x" : 1
} }
});
// If there's no effected row via inc operation, - sub-element doesn't exists at all
// so let's push it
if (inc.value == null) {
date[date.key] = date.value;
var up = db.getCollection(my_coll).update(
{
"_id.x" : 1
},
{
$push : {}
},
{ writeConcern: { w: "majority", wtimeout: 5000 } }
);
// No document found for inserting sub element, let's create it
if (up.nMatched == 0) {
db.getCollection(my_coll).insert({
"_id" : {
"x" : 1
},
"value" : {}
});
}}
Resulting data-structure:
data = {
"_id" : {
"x" : 1,
"y" : 1
},
"value" : {
"2_0" : [
{
"date" : ISODate("2014-12-17T00:00:00.000Z"),
"x" : 1
},
{
"date" : ISODate("2014-12-18T00:00:00.000Z"),
"x" : 2
}
]
}
};
In short i have to apply these actions to process my data:
Try to increment values.
If there's no effected data by increment operation push data to array.
If there's no effected data by push operation create new document.
Problem:
In some cases aggregation result returns more than 2k results i have to apply mentioned steps who causes performance bottleneck. While i'm processing already aggregated data, - new data accumulates for aggregation and later i cannot apply even aggregation to this new raw data because it exceeds 64MB size limit due firsts slowness.
Question:
How i can with this data-structure improve performance when increasing x ( see data-stucture ) values or adding-sub elements?
Also i cannot apply mongodb bulk operations due nested structure using positional parameter.
Maybe chosen data-model is not correct? Or maybe i'm doing not correctly aggregation task at all?
How i can improve aggregated data insertions?